GROWTH OF THE THIN SUFACE FILMS ON LITHIUM IN ALKYL CARBONAT SOLUTION

Liraz Larush, Ella Zinigrad, Yossi Goffer and Doron Aurbach*

Department of Chemistry, BarIlanUniversity,

Ramat Gan,Israel

Abstract. In this paper we report on attempts to use differential scanning calorimetric measurements of aged Li electrodes for the study of the kinetics of the growth of surface films on the active metal. Standard, commonly used alkyl carbonate solutions such as ethylene and di-methyl carbonates (EC-DMC) with LiPF6 were explored. Heating Li samples in solutions after aging by DSC, resulted in well resolved curves of reaction heats vs. temperature. Exothermic reactions occurring at temperatures below 150 ºC could be attributed to changes related to the surface films and their heat evolved, increased as a function of storage time and hence these heats represent the thickness of the surface films which grows upon storage. Scanning electron microscopy (SEM) of the Li surfaces as a function of storage and heating to different temperatures confirmed that the thermal reactions of Li surfaces in these solutions up to 150 ºC relate to the surface films only. XPS studies revealed that these processes of the surface films change the metastable organic Li salts to more stable inorganic compounds such as LiF and Li2O. Massive red-ox reactions between the salt anion and the solvents and between the solution species and the active metal, occur at temperatures higher than 150 ºC. The kinetics of growth of the surface films on Li shows an inverse logarithmic behavior, expected for thin surface films with which the rate limiting step of their growth depends on ions transport across the film.

Introduction

Lithium, lithiated carbon and Li alloy electrodes react spontaneously with all relevant polar aprotic electrolyte solutions to form insoluble products that precipitate as surface films.1 The reduction of most of the commonly-used polar aprotic solvents and salt anions by lithium or lithiated compounds (Li-C, Li alloys) forms Li salts (organic and inorganic), which are insoluble in the parent solutions.2 These surface films are usually electronic insulators, but Li ion conductors, because the property of most of Li salts, which behave like solid electrolytes as thin films. Their electronic/ionic behavior relates to the unique properties of the ionic bonds that Li forms with most of the non-metallic elements.3 The basic scenario of active reducing surfaces in aprotic nonaqueous Li salt solutions is that the surface films formed initially, allow electron tunneling through them until they reach a thickness of several tens of nanometers, what blocks further electron transport from the active metal to the solution through the surface film.4 Thus, a passivation is reached and the surface films stop growing. This process forms protecting surface films that are called solid electrolyte interphase (SEI).5 The surface chemistry and physics of lithium and other reductive lithiated materials have been intensively explored over the last few decades because the surface film phenomena on Li, Li-C, and Li alloys (Li-Si, Li-Sn, etc.) are critical to the performance of most relevant anodes (negative electrodes) in Li and Li ion batteries.6-12 Various surface sensitive techniques have been used in conjunction with electrochemical methods for the intensive studies of the surface phenomena of these reactive electrodes. These included FTIR,13 Raman,14 and AFM.15 The thermal behavior of surface films on graphite electrodes has also been explored in recent years.16,17

Despite the variety of tools that can be employed in the study of surface phenomena on electrodes, a kinetic study of the formation and growth of surface films is very difficult because in situ techniques such as FTIR or Raman suffer from the masking effect of the solution. In situ AFM senses only the outer surface of the electrodes. Ex situ techniques are limited in their ability to provide authentic information on dynamic processes. Electrochemical tools such as impedance spectroscopy can probe the electrodes' surface dynamics. However, the information obtained is not unambiguous, and the interpretation of the data is never straightforward when very active electrodes (such as Li metal or its alloys) are involved.

In this work we attempted to develop a tool for the dynamic study of surface film formation and growth on Li (and lithiated electrodes) based on calorimetry. Storing active electrodes in solution, allowing surface films to grow during certain periods of time, and then applying an in situ calorimetric tool such as DSC, can provide useful information about both the relative status of the surface films at ambient temperatures (at which they grow) and their thermal properties at elevated temperatures. Comparing the relevant heat of reactions measured during temperature scanning after different periods of storage, may provide significant dynamic information. In the present study, Li metal samples storied in commonly-used standard EC-DMC/LiPF6 solutions were studied by differential scanning calorimetry (DSC), x-ray photoelectron spectroscopy (XPS), and scanning electron microscopy (SEM).

Experimental

Li foils (0.3 mm thick) were mechanically cleaned in an argon-filled glove box in order to remove native surface films resulting from Li reactions with the atmospheric components, and were then smoothed by rolling on them a glass rod. A differential scanning calorimeter (DSC) from Mettler Toledo Inc., Model DSC 822, was used for the thermal analysis. For the storage/thermal studies, lithium disks of 3 mm in diameter, with a total surface area of 0.17 cm2, were cut and placed in standard high pressure gold-plated stainless steel crucibles, 30 μl in volume (Mettler Toledo Inc.) in a glove box (VAC, Inc.) under highly pure argon atmosphere. The crucibles were filled with 1M LiPF6/EC:DMC 1:1 solutions, and then sealed. The samples were stored for a while and were then studied by calorimetry at a 40-350 ºC temperature range at a heating/cooling rate 1 ºC/min. The PeakFit Version 4 software (SPSS Inc.) was used for the peaks deconvolution in the thermograms thus obtained. The Pearson Type IV distribution18 was applied for empirically characterize the shape of the DSC data. This data treatment is especially well suited, to model pull distributions, which are asymmetric. The overall thermal effect (i.e. heat evolution) was estimated by the STAReSoftware, Version 8.01. The first exothermic peak obtain in all the experiments, was described well by the Pearson Type IV distribution. In this respect, the first exothermic peak was distinguished from the second one (which appears in all of the thermograms obtained). The shape of the second exotherm did not permit its fitting by only one peak. Therefore, the thermal effect of the second exothermic processes was calculated as the difference between the overall heat evolution and the response of the first exotherm.

Pristine Li foil and Li surfaces after storage and heating were characterized by scanning electron microscopy (SEM, JEOL JSM 840) and XPS. For the microscopic studies, Li samples taken from the crucibles after the processes could be used.

The samples for XPS measurements were 1cm2 square Li foil pieces that were stored at room temperature or stored and heated to 100 or 140 ºC in a 1M LiPF6/EC:DMC 1:1 solution in stainless steel vessels that filled and sealed in a glove box. XPS analysis was performed by a Kratos AXIS-HS spectrometer, using a monochromatized Al Kα source. Survey scans were run at 75 W at PE of 80 eV. High-resolution spectra were acquired at the same power source in a hybrid mode analyzer at detection pass energies of 20 eV. All measurements were carried out with an electron flood gun with low energy electrons for charging neutralization. Prolonged measurements did not reveal any susceptibility to X-ray radiation by the Li samples. All the spectra were calibrated by referencing the binding energy of the lowest C 1s peak to 285 eV.

The air sensitive Li samples were transferred from the glove box to the XPS pre-chamber with no exposure to any atmospheric contaminants, by a homemade transfer device that includes a magnetic manipulator and a gate valve, loaded in the glove box, and then it is attached by a Conflat flange sealed with Viton O-rings, to the XPS pre-chamber pumped down to 1x10-6 Torr. Data processing was carried out with the VISION 2.1 software (“Kratos”). Curve fitting was performed using a 80/20 Gaussian/Lorentzian line shape.

Results

DSC measurements. Li samples were stored at room temperature in 1M LiPF6/EC:DMC solutions during various time periods before measurements. Typical DSC curves of Li samples at a heating/cooling rate of 1ºC/min(Fig. 1) show two consequent exothermic peaks in the 805-14035 ºC range. The peaks are irreversible, as proven by DSC measurements during cooling at the end of the heating process. As can be seen in the curves of figure 1, the exothermic peaks are preceded by small endothermic processes in the range 80-90 ºC (which also increase upon storage), that may relate to processes of the surface films. Since these peaks too small and hence close to the noise level, we could not treat them in the framework of this study.

Both the first and the second exotherms demonstrate increase in the reactions' heat as a function of storage time at RT before the calorimetric measurements, as demonstrated in Fig.1.

The dependence of the heat of the exothermic reactions, (QV,1 and QV,2), on storage time is presented in Fig.2.

It is assumed that the two exothermic processes between 80 and 140 ºC relate only to the surface films. They do not correspond to the same surface compounds but rather reflect a serial conversion of the meta-stable surface species formed at RT to more stable compounds upon heating. The most probable assignment for these processes, observed at the moderately elevated temperatures, are further reduction of meta-stable surface species (e.g. ROCO2Li and ROLi compounds) by Li (due to an increase in the electronic conductivity of the surface films as the temperature arises) and their partial decomposition to form stable species such as LiF and Li2O.

Fig. 1. Typical DSC measurements of Li samples that were stored during different periods of time at room temperature before the measurements in closed crucibles filled by 1M LiPF6 solutions in EC:DMC 1:1: 1-24.5 h, 2-33.0 h, 3- 43.4 h, 4-54.8 h. Heating/cooling rate was 1 ºC /min.

Fig. 2. The dependence of the heat of the first exothermic reaction of the Li samples stored in EC:DMC 1:1 /1M LiPF6 solutions as a function of storage time at RT.

The decomposition of solution components should be assigned to the thermal processes observed at temperatures higher than 150 ºC. Inorganic end-product surface species such as LiF and Li2O should be stable at high temperatures. The meta-stable surface species accumulate in the outer part of the surface films, since in the inner part of the surface films, close to the Li metal, all the species are expected to be in their final, low oxidation state, due to their contact with lithium. Hence, the logical scenario that seems to be well reflected by the DSC curves described herein is that as the temperature rises and the electron tunneling through the surface films is facilitated, all the metastable (organic) compounds of the surface films are reduced. These reduction processes are reflected by the mild exothermic reactions below 150 C. At temperatures above 150 C, pronounced exothermic processes are detected, which should be attributed to both intrinsic red-ox processes of the electrolyte solution19 and the massive reaction between the solution and the lithium, as the passivation is broken at elevated temperatures (especially after Li melting > 180 C). The DSC curves (Fig. 1) also reflect the fact that the surface films on Li are multi-components, as was concluded from previous spectroscopic studies.20,21 Hence the fact that two exothermic peaks are detected below 150 ºC, the second of which is very broad, should be attributed to the multi-component structure of the surface films formed on Li.

Surface studies. Fig. 3 shows a typical SEM micrograph of Li a surface stored, and then heated, in EC-DMC LiPF6 solution to 140 C. A SEM micrograph of a Li surface after storage in solution at RT is also presented in this figure for comparison. The Li surface morphology becomes irregularly cracked (Fig. 3b) due to both storage and heating, but the basic morphology remains somewhat similar. These SEM micrographs show that heating the Li samples in solutions up to 140 C related to processes of the surface films that change their morphology and not to pronounced Li-solution reactions (that produce very rough surfaces).

All the XPS spectra of Li surfaces stored in solutions show Li, F, O, C peaks. XPS spectra show typical C1s and F1s peaks of Li surfaces after storage in EC-DMC/LiPF6 solution at RT and after storage at RT (same period of time) and heating the sample in solution to 100 ºC. It is important to note that all the samples studied were nearly identical. Thereby, the comparison of the peaks' area in the XPS charts reflects true quantitative differences in the content of the element under heating. It is clear that carbon species disappear from the Li surface upon heating while fluorine species become more dominant. These results further show that heating below 140 ºC leads to processes related to changes/decomposion of the surface films and not to massive reactions between the Li and solution species (which should form high amount of carbon containing surface species). The broad F 1s peaks that characterize the x-ray photoelectron spectra of all the Li samples is an evidence of the presence of at least two F-containing surface species in varying proportions. Hence, the F 1s spectra of all the samples were fitted using two peaks. The fitted peaks correlated well with the experimental data. Basically, all the spectra contain one peak at high binding energy (BE), around 686.5-687.7 eV, and one at low BE, around 684.8-685.9 eV. Judging from the NIST data base, one can suggest that the low BE peak belongs to a simple ionic fluoride salt, namely LiF. The high BE F 1s peak can originate from more complex compounds such as LixPFy, LixPOFy, or organo-fluorine compounds. The latter possibility is less plausible, as the C 1s spectra did not reveal features at BE higher then ca. 291 eV, which can be attributed to organo-fluorine compounds. No substantial trend was found regarding the correlation between the sample history and the concentration ratio between the high and low BE F 1s peaks. However, it should be mentioned that the heated samples contained the higher concentration of the low BE F 1s species than that of the high BE species, which suggests that the heated samples surfaces are rich in LiF.

Fig. 3. SEM micrographs of Li surfaces: (a) After mechanical stripping in dry argon atmosphere and then storage in solution at RT (24 h); (b) Li surface prepared as (a) stored in LiPF6/EC:DMC 1:1 solution for 24 h. at room temperature and then was heated to 140°C. The light lines of irregular shape appearing in the images were formed by the mechanical stripping and are clearly seen in both micrographs. The white arrows in 4b, show cracks in the surface films, resulted from heating. A scale appears in each micrograph.

The C 1s spectra from all the samples are complex, and reflect the existence of several carbon-containing species on the Li sample surfaces. The C 1s spectra were all fitted to the smallest number of peaks that gave a good fit, namely 4 peaks, except for the sample stored at RT (no further treatment), which C 1s spectra fitting needed a minimum of five peaks. All these C 1s spectra contained a peak at ca. 290.2-290.5 eV that can be assigned to oxalate or to organic carbonate. The most interesting point that can be learned from these spectra is that the heated samples showed significantly more low BE C 1s features than the samples that weren’t heated. This suggests that the original carbon-based species in the surface films formed at ambient temperatures, such as carbonates and alkoxides (the peaks at around 290 and 286-287 eV respectively), transform at elevated temperatures to carbon species at lower oxidation states. Since the oxygen peaks are poorly resolved, we did not attempt to fit them. In addition, since the positions of the lithium peaks in the leading data bases (XI SDP V2.3XPS International, Kawasaki, Japan, 2001 and the NIST Standard Reference Database 20, Version 3.4) are very scattered, we decided not to try to assign the Li peak to specific compounds.

The most important insight from the XPS data was gained by following the relative concentration of the various elements in the surface films on lithium (Li, O, C, F) as a function of storage and heating. As can be seen in Table 2, there is a dramatic difference between the heated and unheated samples. The most striking change is that the heated samples become very rich in F and Li, compared to the unheated samples, which are richer in carbon.

Hence, these surface studies confirm the above mentioned assumption that heating Li samples leads to a further reduction and decomposition of their surface species, which enrich the Li surface with compounds such as LiF and Li2O at the expense of reductions in the presence of species with carbon-oxygen bonds.

Discussion

The DSC curves of Li samples stored in the electrolyte solutions shows that it may be possible to well separate between thermal processes that reflect changes in the surface films and processes in which massive reactions of the solution species take place. The latter include two type of reactions: red-ox reactions between the salt anion (an oxidizer) and solvent molecules (reducing agents)19 and bulk reactions between the active electrode and solution species, as the passivation is broken down at high enough temperatures (specially intensive as the lithium melts around 180 ºC).

The Li surface studies by SEM and XPS support the conclusion that we can really probe thermal reactions which relate mainly to the surface films alone. Thereby, we used the heat values of the exothermic transformation processes related to the surface films on Li up to 140 C as a function of storage time, in order to estimate the rate of the film growth. The Mott-Cabrera theory predicts an inverse logarithmic law for the growth of very thin oxide films (100-1000Å) on metals under oxygen atmosphere, and for the anodic formation of oxide films in electrolytes containing oxide ions.22 According to this theory, ions and electrons move in the solid ionic films independently. At a low temperature,electron tunneling from the bulk metal to the oxidizing species occurs rapidly compared to the ionic motion across the surface film. The rate-limiting step for mass transport in the oxidation process that forms the films is at the metal-oxide interface and involves the entry of the metal 'ion' from the metal into the oxide lattice.